like tends to mate with like—called assortative mating. This coefficient of resemblance between husband and wife in regard to phthisis is about the same as the correlation of resemblance between husband and wife for eye color, stature, longevity, general health, truthfulness, tone of voice, and many other characters. No one will suppose that life partners "infect" each other in these respects. Certainly no one will claim that a man deliberately selects a wife on the basis of resemblance to himself in these points; but he most certainly does so to some extent unconsciously, as will be described at greater length in Chapter XI. Assortative mating is a well-established fact, and there is every reason to believe that much of the resemblance between husband and wife as regards tuberculosis is due to this fact, and not to infection.8
Again, it is objected that the infection of children is not a family matter, but due to tuberculous cows' milk: how then does it appear equally among the Japanese, where cows are not tuberculous and cow's milk rarely used as an infant food: or among such people as the Esquimaux and Polynesians, who have never